Field notes
1,406 notes from my AI research vault, mapped by topic. Pick a topic to see its notes, or search.
Click the web, then scroll to zoom
Every note is in the list below. The web of topics draws here once the page has loaded.
1,406 notes and 240 topics.
- Question answered
- Source note
- Topic
- Shared notes
All notes, newest first
1,406 notes, page 28 of 59
Generalized Kalman filter based temporal difference reinforcement learning
The paper introduces a new reinforcement learning framework that combines conditional expectation theory with a generalized Kalman filter. This approach allows for the estimation of both the value function and its…
Hard Guarantees at a Measured Price: Entropy-Stable Learned Finite Volumes for Compressible Flow
The author presents a new learned finite volume scheme for two-dimensional compressible flow that guarantees physical admissibility by construction and has an entropy-stable interior flux. The scheme was evaluated…
HeadCast: Casting Attention Heads for Efficient Autoregressive Video Generation
Researchers propose HeadCast, a framework that improves the efficiency of autoregressive video diffusion models without requiring re-training. The framework categorizes attention heads into four types based on their…
HijackKV: New Threat in Position-Independent KV Cache Reuse
Researchers have identified a new threat called KV Cache Hijacking in position-independent KV cache reuse, which is used to reduce inference latency in large language models. This threat allows an attacker to silently…
Instance Hardness-Based Relevance for Imbalanced Regression
A new approach called Instance Hardness-based relevance function (InHaR) is proposed to identify rare instances in regression problems, particularly useful for complex scenarios like bimodal distributions where…
Interpretable Fuzzy Rule-Based Regression Extension for Ex-Fuzzy Library
Researchers present an extension for the Ex-Fuzzy library that makes Mamdani-style fuzzy regression more accessible in modern machine learning software. This extension allows for learning from data and offers…
Label-Free Finite-Volume-Residual Training of Attention Graph Neural Networks for Coupled Thermo-Fluid Fields
Researchers propose a new method to train Attention Graph Neural Networks for predicting three-dimensional thermo-fluid fields. Instead of using conventional numerical solvers that incur high computational costs, they…
Lipschitzian SLLNs for random functions
The authors present a new Strong Law of Large Numbers (SLLN) for locally Lipschitz functions in the context of the Lipschitz pseudometric. The SLLNs hold under either a topological or model-theoretic condition, and can…
Local Stability and Gaussian Smoothing of Quantized Neural Networks
The authors investigate using Gaussian averaging as a smooth alternative for quantized neural networks. They provide bounds on the difference between the original function (f) and the smoothed version (g), linking it…
Multi-modal transformer for signal classification in nanopore blockade experiments
Researchers present a deep learning architecture for analyzing complex signals from nanopore blockade experiments. The model, which combines multiple signal representations, outperforms existing methods on a peptide…
Multi-stage Dynamic Selection for Cross-Project Defect Prediction
The paper proposes a new framework for Cross-Project Defect Prediction (CPDP) that addresses distribution shift issues between training and target projects by using a two-stage multiple classifier system selection…
Non--negative matrix factorization using the extit{R} package extsf{nnmf}
This research presents a new R package for Non-negative Matrix Factorization (NMF) and compares it with two other widely used NMF packages in R. The comparison is based on real-world data instead of simulated datasets…
OLEDLM: A Unified Language Model for OLED Molecular Design
Researchers propose a new language model for designing Organic Light-Emitting Diode (OLED) materials. The model generates SMILES sequences of OLED molecules based on specific optoelectronic properties, using a…
On Optimization Complexity of Second-Order Certified Unlearning
The authors study the complexity of removing memorized training data from a machine learning model while maintaining accuracy, a process known as certified unlearning. They present new bounds on the distance between…
PhaseAware: Interpretable Human-in-the-Loop Rehabilitation Scoring with Boundary Monitoring
The study introduces PhaseAware, a framework designed for continuous evaluation of rehabilitation quality during physical therapy sessions. It combines temporal data with phase-specific and body part descriptors to…
PIER: Physics-Informed Environmental Retrieval for Time-Series Modeling
The researchers propose Physics-Informed Environmental Retrieval (PIER), a model-agnostic framework to improve the accuracy of environmental system modeling. The framework is designed to address challenges in limited…
Plausibility-Driven Prioritization of Candidate Biomedical Annotations
The paper presents a framework using biomedical knowledge graphs (bioKGs) to estimate the plausibility of candidate biomedical annotations, aiming to guide experts in prioritizing them for manual curation. The approach…
Statistical Inference for Rank Allocation in Low-Rank Adaptation
In this paper, the authors propose StatLoRA, a statistical inference-based rank allocation method for low-rank adaptation (LoRA) in large language models. The proposed method uses estimated p-values to determine which…
Taming the Security-Energy Paradox: A Green AI Approach to Optimized Android Malware Detection
Researchers have developed an energy-efficient method for Android malware detection using deep learning models, aiming to balance performance and battery life. They used quantized neural networks (QNNs) with reduced…
Test Case Prioritization for DNNs via Neural Collapse Instability
A new method, Neural-Collapse-Inspired Prioritization (NCIP), is proposed to improve test case prioritization for deep neural networks (DNNs) in safety-critical domains where testing budgets are limited. The method…
The Blessing of Dimensionality: How Near-Orthogonality in High-Dimensional Spaces Explains Temporal Portability
The paper addresses two open questions regarding a method called PortLLM, which adapts large language models after continual pretraining. The authors perform an extensive empirical study to explore the long-term…
Time Series Network Utilization KPI Forecasting Using Advanced AI/ML Models
This study evaluates various AI/ML models for forecasting bandwidth utilization in networks, aiming to improve capacity planning and reduce costly over-provisioning.
Towards Miniature Humanoid Tele-Loco-Manipulation Using Virtual Reality and Reinforcement Learning
The paper describes a new control stack for miniature humanoids that allows a single operator to see, manipulate, and navigate in distant environments using Virtual Reality (VR) for upper-body teleoperation and…
Variance-reduced Domain Adaptation using Paired Sampling
The paper proposes Paired Sampling for Domain Adaptation (PSDA), a new stochastic variance reduction technique designed for distribution-matching frameworks in unsupervised domain adaptation (UDA) that suffer from high…
An AI research agent drafts these notes from public papers and posts, and each one keeps its source link. Only the folders I choose to publish leave my Obsidian vault.
